"""Склейка набора файлов ответа ИС в один ответ (spec image-recognition-gateway «Сырой ответ…», D-4, R-24).

`merge_answers(job_id, files)` разбирает каждый файл `json.loads` независимо: не JSON — `response = null` и
`parse_error`, задание не падает; файлы — по возрастанию имени. `dump_merged` — байты объекта `raw/{job_id}.json`.
"""

import json
import shutil
from pathlib import Path

import pytest

from src.gateway.hotfolder import AnswerFile
from src.gateway.merge import dump_merged, merge_answers

API_DIR = Path(__file__).resolve().parents[2]
REAL_FIXTURES = API_DIR.parent.parent / "fixtures" / "ocr"
JOB = "3f1c2b7a-0000-4000-8000-000000000001"


def test_structure_for_one_file() -> None:
    merged = merge_answers(JOB, [AnswerFile(f"{JOB}.json", b'{"file_uid": "x", "error": {"code": "", "message": ""}}')])
    assert merged == {
        "job_id": JOB,
        "files": [
            {
                "name": f"{JOB}.json",
                "response": {"file_uid": "x", "error": {"code": "", "message": ""}},
                "parse_error": None,
            }
        ],
    }


def test_files_are_ordered_by_name() -> None:
    files = [
        AnswerFile(f"{JOB}_ofr.json", b'{"n": 2}'),
        AnswerFile("0.json", b'{"n": 0}'),
        AnswerFile(f"{JOB}.json", b'{"n": 1}'),
    ]
    merged = merge_answers(JOB, files)
    assert [f["name"] for f in merged["files"]] == ["0.json", f"{JOB}.json", f"{JOB}_ofr.json"]
    assert [f["response"] for f in merged["files"]] == [{"n": 0}, {"n": 1}, {"n": 2}]


@pytest.mark.parametrize(
    "data",
    [b"", b"not json", b'{"a": ', b"\xff\xfe<html>", b"[" * 100_000],
    ids=["empty", "text", "truncated", "bad_utf8", "deep"],
)
def test_invalid_json_becomes_parse_error_and_others_are_parsed(data: bytes) -> None:
    files = [AnswerFile(f"{JOB}.json", b'{"ok": true}'), AnswerFile(f"{JOB}_ofr.json", data)]
    merged = merge_answers(JOB, files)
    first, second = merged["files"]
    assert first == {"name": f"{JOB}.json", "response": {"ok": True}, "parse_error": None}
    assert second["name"] == f"{JOB}_ofr.json"
    assert second["response"] is None
    assert isinstance(second["parse_error"], str)
    assert second["parse_error"]


@pytest.mark.parametrize("data", [b"[]", b'"text"', b"42", b"null"])
def test_valid_json_of_any_root_is_kept_as_is(data: bytes) -> None:
    """Корень не объект — это забота нормализатора (`INVALID_FORMAT` файла), склейка хранит разобранное как есть."""
    [entry] = merge_answers(JOB, [AnswerFile(f"{JOB}.json", data)])["files"]
    assert entry["response"] == json.loads(data)
    assert entry["parse_error"] is None


def test_empty_set_gives_empty_files() -> None:
    assert merge_answers(JOB, []) == {"job_id": JOB, "files": []}


def test_dump_is_utf8_json_that_round_trips() -> None:
    merged = merge_answers(JOB, [AnswerFile(f"{JOB}.json", '{"m": "Не удалось"}'.encode())])
    data = dump_merged(merged)
    assert "Не удалось".encode() in data  # кириллица читаема в S3, не \\uXXXX
    assert json.loads(data) == merged


def test_dump_with_lone_surrogate_is_valid_utf8() -> None:
    """`\\ud800` в ответе ИС разбирается в одиночный суррогат: объект всё равно записывается валидным UTF-8."""
    merged = merge_answers(JOB, [AnswerFile(f"{JOB}.json", b'{"m": "bad \\ud800 tail"}')])
    data = dump_merged(merged)
    data.decode("utf-8")
    assert json.loads(data) == merged


@pytest.mark.fixtures
def test_real_fixtures_are_merged(tmp_path: Path) -> None:
    if not REAL_FIXTURES.is_dir():
        pytest.skip("fixtures недоступен: положите репозиторий fixtures рядом с workspace")
    for name in ("aep_3.json", "aep_3_ofr.json"):
        shutil.copy(REAL_FIXTURES / name, tmp_path / name)
    files = [AnswerFile(p.name, p.read_bytes()) for p in sorted(tmp_path.iterdir())]
    merged = merge_answers(JOB, files)
    assert [f["name"] for f in merged["files"]] == ["aep_3.json", "aep_3_ofr.json"]
    assert all(f["parse_error"] is None for f in merged["files"])
    assert merged["files"][0]["response"]["error"] == {"code": "", "message": ""}
    assert merged["files"][1]["response"]["error"]["code"] == "INVALID_DOCUMENT"
    assert json.loads(dump_merged(merged)) == merged